Notes from Portand ArcSDE User Group Meeting

Report by Theresa Valentine, Corvallis Forest Science Lab
I attended a meeting in Portland on Friday where John Sharrard from ESRI did a demo and talked about topology in ArcGIS 8.3. I thought I'd share some of the information I picked up at this meeting.

Topology lets you set up relationships between different layers of data. Note, these layers have to be in a single Feature Data Set (all with the same projection). It works in the personal geodatabase and SDE geodatabases.

It is important to remember that data can move if you give it topology. (works similar to a clean in coverages). You have to set up your cluster tolerance to keep movement to a minimum. You can rank movement (one layer is ranked 1 and anther ranked 2...the ranked 2 would move toward the rank 1)..however, there is no unranked (do not ever move this layer, so even you most stable layer could move...again, give it a very very small tolerance). You can have up to 50 ranks per feature dataset.

You also can select or define rules for the relationships. There is a poster you can print out on the Desktop CD that shows all the rules and has pictures to help explain them. The rules travel with the geodatabase. They also are documented in the metadata. These are where the real power of topology come in. For example, you can set 'no dangles' as a rule, then when you validate, you get a list of errors, and you can fix them or call them exceptions.

Validating creates an additional feature class (topography) and this is where the snapping occurs. You have to be careful when you do the validate, (at least the first time) because it can be very CPU intensive. It's a client program, so if you are using SDE, the work is done on your own computer, not on the machine with SDE. It might be something you would do overnight. We were told that the snapping occurs on the first validation, and that you should not see data shifts after every validation you do (as compared to a clean, that when you do it many times, your data starts to drift around). Also, arcs don't automatically intersect when you do a validate. John was going to show us something about that, but ran out of time.

You can only have one set of topology per feature dataset.

Some issues that came up within the group discussion were
1. that you also have one set of permission per dataset, which could be a problem. For example, if you had a dataset on parcel lines that you didn't want to move and didn't want anyone to edit, but you wanted them to have topology with zoning, taxlots, and perhaps voting districts, you would have to give anyone editing zoning, taxlots, or voting districts, access to edit the parcel lines. Many organizations split up these functions, so this might cause a problem.
2. You need to really look at your data base design and work flow if you implement topology. You need to figure out what data layers to put into each feature data set, and you have to decide what the hierarchy of the layers within that dataset are.